 Hello everyone and welcome to the Latin American webinars on physics My name is Nicolás Bernal from the ICTP safer in Sao Paulo, and I will be your host today So our speaker today is Mattia for NASA From the Grappa Institute in Amsterdam. We'll talk about the refuse gamma ray background So Mattia Receive his PhD from the University of Padova Right. Yes, correct. Yeah, so and after a couple of postdocs so two appointments one in the Institute of Astrophysics and Alucía and Then the University of Nottingham He moved to Amsterdam where he's a postdoc in the in the Grappa Institute So Mattia stole today's title the anisotropies in the diffuse gamma ray background And we are glad to to have him here today as our as our speakers So first let me remind that you can be part of the discussion writing question and comments using go Google Q&A system and on Twitter with the hashtag LAWP so let's in American webinar some physics So Mattia, I think you can Yeah, thanks, Nicolás. I guess you can hear me, right? Yeah, good. So well, I'm very happy to take part of this series of webinars I think it's a very good idea. So thanks for the opportunity and I'm also quite happy that I can talk about one of my favorite topics, which is the diffuse gamma ray background. So I think I can maybe show you the slides Let's see Let me put them on full screen. Tell me if there's any problem. You see them, right? It's black. I think it's black. Good Let me try again. It's not full screen, but I mean Okay, let me try you see them all full screen now No One last try it's black again. I think you can you can do it without the full screen But now you don't see even the slides now it's completely black for me at least Sorry about this Can you see them now? Yes, yes Let's go from here. So as I was saying, this is about the diffuse gamma ray background It's I'm gonna mainly tell you about a project that I conducted with all of these people and the paper is finally out it's been published on the archive since Three or four weeks. You see there the the number and it's a it's a measurement of the Anisotropy angle of power spectrum of the diffuse gamma ray background So my idea is to tell you something about the diffuse gamma ray background then report there's also our measurement and Let's see how I go with time but probably I will end up mentioning dark matter only towards the very end of the presentation Even if that matter is actually the reason why I Got interested in the topic with diffuse gamma ray background at the beginning So let me start This is the this is a map of the gamma rays skies sky as the gamma ray telescope Fermilac sees it He actually contains all the gamma ray events observed by Fermi after five years of data taking Above an energy of one gd the map is in galactic coordinates. So the electric center is at the very center and this horizontal Band that you see is the emission associated with our own galaxy with the milky way That's what what I will call the the galactic foreground due to the interaction of cosmic rays with interstellar medium and Interstellar radiation fields of the milky way and then all over the map. You see a lot of these very bright spots points. These are individual point sources that Fermi detected In the most recent version of its source catalog, what is called the 3fgl the catalogue contains almost 3000 sources and a huge effort has been going on in trying to Identify those sources. Some of them as have been identified as blaze us. So this is a mission produced by a Supermassive black hole when it shoots is light directly towards us Some other of these sources can be other galaxies like the milky way, but of course for from outside Our own galaxy or you can have other artificial objects like pulsars or what I call to misaligned AGNs the the thing is that if you start from this map and start to Peel away to subtract the components that that we know of so the galactic Foreground and the sources well, you don't you don't end up with zero There's a residual emission that comes from approximately all over the direction of the sky in the Neonomogeneous way, and that's what I will call the diffuse camera in background. It's residual Emission and it's not something that you should that should surprise you Because I mean at the end of the day this sources the one that you see here are only those that emit With the sensitivity without without the flat survey, which is higher than the sensitivity of Fermilat But of course there are other objects that they meet in a fainter way that are weaker and that Fermilat will not be able to Individually detect individually observe those objects will still emit light and cumulatively they will produce a somehow diffuse glow that will come from all the directions around us And that's a diffuse gamma-ray background the cumulative emission of all the sources. They are not bright enough to be detected individually If I go to the following slides you see here the Energy spectrum diffuse gamma-ray background. So this is flux on the y-axis So number of photons third units of collective collective area observation time energy and and Stay radiant so flux as a function of the of the energy you should look at this I could for example at the black data points This tells you basically the intensity of the fusicam background as a function of the energy and how I wear these points Better mind. Well, what you can do is build a model for The whole gamma-ray emission that we saw in the previous slides in that in the map So your model will have to contain a collective a component for the collective if using mission a component for each one of the individual sources, and then you also add a Template for the diffuse gamma-ray background and you leave the normalizations in front of all of these terms Free to vary and use the model to fit the data that we saw in the map earlier the best fit normalization of The template that you use for the shoes gamma-ray background will give you the intensity the measured intensity of the flux And we do you do that for each energy beam and you get this this points over there Actually, there are not just one data set You see three different datasets the blue the black and the right one and that's because depending on how you model The diffusion of the the emission of the Miki way you end up with slightly different values of the diffuse gamma-ray background overall this Systematic uncertainty can be as large as to 15 30 percent depending on the on the energy range And that's expressed in the plot by the yellow band Anyway, overall, you can say that the spectrum of the fuse gamma-ray background is Compatible with being a power low with a slope of around 2.32 At least a lower energies and then at around 120 GV There's an exponential cutoff that explains the iron edge part This is also the a good point to tell you that the community does not agree on a way to Call this emission so I've been I've been referring to it as the diffuse gamma-ray background But other people call it the isotropic gamma-ray background. So IGRB. That's why it's called That's what is written in the figure. Other other people call it the extra galactic gamma-ray background. So EGV I'm telling you this simply if if you Find yourself reading a paper where they use different names One interesting thing is that you can you can actually You know estimate how much different classes of unresolved astrophysical sources Contribute to this diffused gamma-ray emission and we can see that in this other plot So the black dots the black data points again show the same data set as before and then there are these four different colored bands Bands this this this bands they show how much possible Classes of astrophysical optics contribute or accounts for this diffused gamma-ray background So you have the blazer that I was mentioning before in in orange The other galaxies in blue, middle and the AGM in green and the possibility of you also put stars down here you have Bands and not just one line because estimating these contributions involve some kind of uncertainty So you end up with a with a band and How can you do this? How can you? try to Measure or estimate how much different of specific astrophysical sources contribute to the emission Well, if you think for example at blazers These is a class of sources that dominate the catalog of the object detected by Fermi almost one-third of all the seriously detected by Fermi in the in the last catalog in the three FGL catalog are Blazers and and then you can you can build plots like this one where you basically see the number of blazers detected Above a certain flux as a function of the flux. So you start a large fluxes where you have Not many objects and then you go down you have more and more blazers until you reach the sensitivity of your telescope And then you can fit these points and you see for example that a broken power low represents Provides a good fit to this data If you assume that this same behavior this same broken power low also extends below the threshold of your telescope Then what you can do is Integrate the area below the gray line And that will give you an estimate of how many sources Will remain unresolved and how much they they contribute maybe a bit light Of course, this is a very good idea for for blazers because you have a lot of them But if you focus on another class of sources like for example star forming galaxies Well, in that case you only have Very few sources detected by fairly we're talking about of the order of of 10 And you cannot do population study. You cannot build plots like this one with only 10 sources So in that case what you can do is try to identify some some relations some correlations with the luminosities of star forming galaxies in gamma rays With respect to the lunacy in other frequencies in other wavelengths for example in the case of of Um Star forming galaxies we use infrared data and this is what you see there in this in this other plot So the the the black data points are the few Star forming galaxies detected by Fermi and they relate the these plots related the most in gamma ray with the one in infrared You see that those points fall more or less along the same the same line and at lower frequency The the the sample of results star forming galaxies is much larger than in gamma rays So you can do population studies at lower frequencies and then you use this Established relationship to translate whatever you have learned at lower frequencies to the gamma rays And there you can you can estimate how much star forming galaxies contribute to the diffuse gamma ray Anyway, if I go back one slide, it doesn't matter how we get to This color bands what you can see already from this plot is that these three classes of sources So blazers galaxies and mesoline hgm They seem to do a very good job in reproducing the data and reproducing the black data points Which tells you that If you if you come up with any other possible Emission mechanism any other possible contribution You know for certain that that contribution will have to be subdominant And you can actually use these data to put some limits on that additional contribution This is what is done. For example in the case of of dark matter. We know that dark matter and halos Surround and bad galaxies and groups of galaxies and those dark matter halos can emit gamma rays gamma rays When dark matter particles annihilate or decay, however, they are normally very faint sources. We haven't been able to uniquely Detect a dark matter induced gamma ray flux So all of that matter halos emitting gamma rays will contribute somehow To the diffuse gamma ray background But we know already from from this plot that that dark matter contribution will have to be Small and you can put some constraints on that precisely from this data Um, good. So we already learned a lot. We know what the diffuse gamma ray background is We know that you can estimate How much different hostages contributed to it and how you can use that to put constraints on other components And this is of course not the the end of the story And I mean if you think about it until now, we've only talked we've only discussed um, the the overall average emission of the diffuse gamma ray background But you may it is possible that you can learn A lot more about that if you if you if you think if you focus on this morphology on how much the emission Changes from from point to point and one standard way of estimating that is computing the what is called the angular power spectrum This is exactly the same technique that is That proved to be very efficient in the case of the cosmic microwave background What you do is you you start from an emission And you decompose that in spherical harmonics that will give you the a l m coefficient And the angular power spectrum the c l is the average of the a l m squared with the same l so What what The research program what we would like to have is very similar to what we discuss until now in the case in the case of the of the intensity We would like to have some data on the angular power spectrum And then also at the same time Develop a model that tells you how much different objects different classes of objects Contribute to the angular power spectrum in that way you can see if As it was happening with the intensity you can reproduce your data on developer spectrum in terms of One to few cluster sources, and if you manage to do that, then again you can use your data to constrain other possible subdominant components This is my my research program what I have in mind what we would like to achieve And this this is also one that I will tell you in the next In the next minutes and that and the remaining part of my talk but before entering this I would just want to stress one point that will be important later And this is related to what is called the bosonian noise So if you focus for example in one specific patch of the sky like for example this box over here And then you receive some gamma ray emission from that. This is uh, for example a a fraction a certain patch of the diffuser gamma ray emission I told you that the diffuser gamma ray emission is due to the cumulative radiation of unresolved So in principle, I'm not allowed to put, you know, identify Every source as a as a purple points because these sources are unresolved But this is just a very naive Sketch just just bear with me in a second. I would just want to make a point Um What we would like to do is computing the envelope spectrum on this patch of the sky and use that observable to Get some information to constrain somehow the sources producing the radiation And you may be able to do that or not And that depends a lot on how many sources you have in your region of observation You may be able to do it in this case And indeed you see that the the searches are somehow clustered around a particular region But it may also be the case that your emission is due to Much fewer searches like in this case and of course in this case in the middle There's no hope that you will learn anything about their distribution If you compute the envelope spectrum in this case what you will end up with Is simply something that is flat in l flat in angular multiple or what is also called Poissonian which is a characteristic a Poissonian angle of perspective is a characteristic of a Of sources that are unclustered that are randomly distributed in the sky and this is a a Unavoidable component you will always have a Poissonian component in your signal And if you want to extract information on the distribution of sources, you're better hope that the That term the one that relates to the morphology to how they are distributed It needs to be larger than the Poissonian term Just to give you an idea on the last plot here on the right You see what you expect from the emission due to unresolved blazers. This is a very old Estimate But it's one of the first papers that that computed these terms I wanted to give credit to it What they're doing is plotting the cl the angle of power spectrum as a function of l of multiple And it's also being multiplied by l times l plus one So something that is Poissonian this flat in l appears here as proportional to l square more or less And that's the dotted Red line over here Well, on the other hand the term that actually is relating to how the sources are distributed in the sky is either The purple or the blue lines the dashed line over there And Well, that depends on which one which line depends on the specific of how the blazers are modeled But the important point that I would like to stress is that at least Over here at large multiples the Poissonian term is always larger than the one Arrived to the correlation of searches in the sky. So down here You will never be able to learn Any information about how the search is distributed because you're seeing that you will always be dominated by the Poissonian the Poissonian term This is something that you cannot Change. That's just how how many searches you have in your signal There's nothing you can do about that So just remember this for the following slides in the case of unresolved blazers The angle of their angle of perspective is dominated by the Poissonian Component related to how many searches you have in the sky And then just one last comment When you go back to your data with your map of gamma ray events in the sky and you try to compute the angle of spectrum on that map well In that the case you're talking about, you know gamma ray events detected by Fermin by your telescope and those events are nothing more than a huge list of Gamma rays with a specific energy in a specific direction in the sky when you plug them in a map Again, you're talking about, you know points in a map You they will be associated to another source of Poissonian noise just as this one in this case They will be it will be related to the number of gamma ray events detected and not number of sources But that's another Poissonian term Just as this one unavoidable. So if you want to detect any Significant angle of perspective You you better hope that that signal is larger than the Poissonian term Or associated to them to the number of events in order to distinguish this kind of Poissonian term to the one To the other one the one related to the searches They called the one related to the photons just photon counts photon noise. Sorry So you better hope that your signal is larger than than the photon noise But but one thing that you can do to reduce your photon noise in this case Is simply leave in your telescope letting your telescope on for more and more time You will you will gather and collect more data increase the number of events and reducing the photon counts Okay, so after this introduction, let me tell you about our measurement This is not the first measurement of the angle of perspective. The original and the first one is dated 2012 And it was produced using 22 months of fermion data We now are using almost four times the statistics 81 months We are extending the energy range consider going from the energy window between 150 gv to Three orders of magnitude covering the range between half a gv to half a tv And you see the data that we used in one specific energy beam in the map over there We we computed the upper spectrum Um on the masked sky So we cover we mask all the regions in the sky where We know the diffuse gamma background is just a subliminal component So we must the electric plane and all the sources We actually considered two different masks in one case. We must all the sources in the Latest version of the Fermi catalog of the tactic sources So the all the masking the three of the sources in the 3mgl And then independently we also consider another master only covers their sources present in the older older version of the catalog the 2fgl And in everything that you will see in the next slide the results will be presented for both choices of the mask um What you see Here in this slide is actually our anglophone spectrum estimator. So this Part here is what comes out from the decomposition in spherical harmonics And to that we subtract the photon noise that i was telling you about That is this unavoidable component related to how many gamma ray events your telescope has detected. So A a a significant signal means Something that is larger than cm. So we are looking for excesses over cm when this difference is Significantly different than zero. So that's our Detected signal and then we also have to correct for the effect of the points spread function of the telescope So the angular resolution and we do that by Dividing for this function called wind of infunction wl You see some example of the wls in this photo there. So that's wind of infunction as a function of angular multiple l Different colors stands for different reference energy and you see that the effect of having a finite Angular resolution is that the the the signal wl is dumped decreases At at large multiples if you go at very low energies like in the case of the black curves this damping Already starts to occur at a multiple of around 100 200 And then yes, if you tell you also something about these indexes over there i and j they label The energy beams because not only we compute the angular power spectrum Inside each of the 13 energy beams that we consider So that would be when i is equal to j but for the first time we also compute the cross correlation between the emission Between different energy beams. So that would be the case when i is different from j So in from now on whatever i'm plotting cl whatever you will see the cl is always This angular power spectrum estimator You see some example two examples in this slide. So that's again cl as a function of l on the left For one energy being at low energies and on the right for one energy being at higher energies There are two different data sets a blue and a red one The red one is for the mask covering sources in the three fgl and the blue one When you when we consider the mask around the sources in the two fgl The blue data sets are systematically higher than the red ones and that's because In that case, you're still sensitive to the news order be produced by those sources that will be detected in the three fgl But they are still unmasked in the case of the two fgl And they also both data sets are also systematically Higher larger than zero at least in the case in the in the left panel, which means Uh a detection something a significant detection is a detection of significant power And you should also note those two vertical lines Dashed gray lines that defines our Multiple region so we neglect anything below l of 49 and above l of 706 and that's because we want to We We want to get rid of any possible contamination due to the galactic foreground and also to the Multiple region multiple range where the correction due to the psf becomes too too extreme So the only focus In this central region. We take this remaining 10 data points and fit them with a flat line with a Poissonian angular power spectrum and we we find that in all The 13 energy beams of the autocorrelation and also in all the possible independent combination of energy beams A poissonian angular power spectrum a flat line is always a good description to the data And you can actually see that in these plots the the horizontal solid red line is the best fit Poissonian angular power spectrum to the red circles And the dashed blue line is the best fit poissonian angular power spectrum to the blue squares And the fact that these data are well Described well fitted by poissonian angular power spectrum by a flat line is a good news for me because now I don't have to show you all of these plots for all the different energy beams I can just give you one number which is the intensity where to put this horizontal line So that's what i'm doing in the following slides What you see here is the best fit Poissonian angular power spectrum for the autocorrelation In the 13 energy beams as a functional energy For the mask around the 3 fgl in red and the mask around the 2 fgl searches in blue This is also what is called the anisotropy energy spectrum And well first of all it gives an idea of the significance of our detection. So points around here They have they're different from zero. So signal With a significance of around seven and eight sigma And you go below three sigma only for this last Four points for red points and these two over there. So this is a That the significance of the detection is much larger than the original 2012 measurement And also this quantity The anisotropy energy spectrum turned out to be very very informative to infer to constrain the The the components of the of the anisotropy power of the anisotropy angular power spectrum And that's because you should you should remember that at the end of the day this Power the cl is nothing more than the average of the alm square which comes from a the composition of the original intensity map So if your intensity map depends on energy as it does we know it does The alm coefficient will also depend on on energy and the angular power spectrum will Inherit the same dependent on energy but squared, of course You can actually factor out the dependence on the energy in this way So the basket Poisson angular power spectrum in the combination ij of energy beings Can be written as the intensity in the two energy beings means i and v and j times in energy independent component And if your if your signal is due to only one population of resources Well, that's the end of the story this part the left hand side you measure This is what we are showing here and from this you can directly read you can directly infer how your sources emit in energy Um in the case of more than one population of sources then things become a little bit more complicated The the the contribution of the two populations sum Linearly so you have the case of population a plus the case of population b and inside the population of sources you can do the same The composition as we were doing before but in this case you you you may think of a scenario in which you have For example population a Dominating your signal to low energies with a very soft so steep energy dependence and that is some point Possibly population b will start to kick in with a harder energy spectrum so that the overall Energy dependence of the signal will not be flat will have some feature or breaks And it's just to tell you that indeed any feature that you can detect In the way the cp depends on energy Can be considered as an indication of multiple population or sources contributing to your signal And indeed we were very happy when we saw for the first time this plot Where it does seem that the Under to be energy spectrum is not it's not just flat just a boring power load um, we can be even more Precise if we consider if we take into account also the cross correlation so The cross correlation is I mean we could I could have shown the same plots like this one for the cross correlation as a function of the energy But I think a better way of understanding this is by considering what is called the cross correlation coefficient the r i j Which is simply, you know the cross correlation in the combination of energy beings i j divided By the autocorrelation in the two energy beings separately so if you Keep track of the energy dependence as we were doing before so let's say We have the same expression as before and you put this At the numerator in the case of only one population or service is one that's easy just rewrite the numerator in this way and in the case of the numerator you have i equal to j and you can easily understand by yourself that the energy dependence Exactly cancels from numerator and denominator. So the cross correlation coefficient r will be identically equal to one That's the case for one population or service if you have more than one population or service to continue to your signal than the components some and this um cancellation from the relative to the numerator does not happen anymore, which means that If you have some coefficients different than one, that's again a signal of more than one service classes contributing to your signal And we can see that in in those two plots In the slide so on the left, it's the case where you mask the sources in the 2 fgl And in the right where you must associate the 3 fgl and plotting the cross correlation coefficient Of all the 13 energy beings against the other energy beings. So these plots are by construction symmetric along the diagonal and they're That the diagonal entries are identically equal to one But you see clearly by either there's a region around here when you cross correlating low energy beings with low energy beings where the coefficients are around one But then another region down there when you cross correlating low energy beings with higher energy beings Where the the coefficients are around zero And that's that's more or less falls in the same scenario that I was Depicting before right a population of those services contributed a low energy in a different one contributing that higher This is for 2 fgl in the case of 3 fgl You kind of see the same structure. You have to maybe squint right a little bit Of course, these data points are affected by a larger error bar. So it's difficult to see it um We can even be More Quantitative and what we did was try to fit this data both the auto correlation and the cross correlation By some phenomenological models where we assume one or more search classes And also we assume that those those classes emit either as power loss in energies or broken power loss So the the best fit case in the case of only one population of a service is Planted in the in the panel on the on the on the left by means of the pink and blue line and in this case Whether it's a broken power law or a power law Those feet are not very good ones. You see that both cases underproduce the very two energy beams if you go to two pressure resources, then you need to look at the plot on the right um Here we have the case where there are two power laws in yellow one broken powers and one power law in green But the best description to the data is provided by two broken power laws. That's the thick black line that is That is positive that is shown in both panels both on the right and the left And in this case the the thick black line The two population resources are shown independently by the by the dashed blue lines So again, as I was saying before you have one class We're producing dominating the data Around one gd or below and then the second one which are with a much Flatter harder spectrum Describing reproducing your data at higher energies And in the following slide, I'm giving you the details of the best fit Model parameters of the two political services. We can actually by performing a a likelihood ratio test exclude The 95 percent confidence level the interpretation in terms of only one population resources Just back to the one with two classes. And that's the first time that we can do that So This concludes more or less my uh presentation of the measurement. Of course, this is a very you know naive exercise what we did with the phenomenological description of of of Possible source classes contributed to your signal. What one should do and could do is You know devising or build up a model where you have physically motivated, you know physical searches and and then Use these data to to constrain the model of your Yeah, first principle physically motivated as through physical searches like blazer or or start forming galaxies We're not doing that in in the papers There's still room for for for some work Uh in the paper the the the thing that we do is instead using these data To put constraints to put upper limit on a possible dark matter components to the to the result of this signal And that will be very very brief on this Just telling you that Well, if you want to compare This data with the possible dark matter induced agglomerate spectrum. And of course you have to estimate How the dark matter signal uh looks like you can do that um including components that come from extragalactic matter of halos objects Like you see in the top panels and the galaxy itself in the bottom panel panel You have both the case of annihilation and decay you compute the angular power spectrum on these maps That's what you find so again cl in more specific energy beam As a function of l you see different color bands because we We identify we consider different components to the signal independently And we also take very good care in estimating How much this each component is affected by some Systematic uncertainty relative to how we model the back matter distribution And then you have to compare this prediction to the data, which in this case is By this this black line And and then of course you can use this comparison to put some Exclusion limits on the dark matter signal the very the easiest thing to do will be requiring that there's no energy beam In which the dark matter signal Is larger than the the technique signal than the data And that will give you some exclusion limits that can be translated into upper limits on the annihilation cross-section or environmental particle as a function of mass so everything above this Color line is excluded and we have three lines the blue the black and the red ones Because we we consider these three cases as three possible benchmark related to how we model the distribution of dark matter in the sky Alternatively in the case of of decay in dark matter, then while you have our lower limits excluding everything below This colored line and you should compare this exclusion limit with what is available in the literature. So if I focus on annihilation Our exclusion limit should be compared to the thermal value of the cross-section So dashed gray line here Or to the exclusion limits that you can have from the observation of those ferroidon, which is the Those dashed gray line, you know, fortunately, we are something like a factor of 20 above this other upper limits Or then you can do something more more constraining For example, devising this two component fit to the data where you model at the same time impossible that matter signal plus A postion component, which is supposed to describe, you know, any other astrophysical contribution And in this case, the upper limit comes by requiring that adding the dark matter component do not spoil the Poissonian interpretation of data too much and again Everything above this three colored line or below this band is excluded Thanks to these more aggressive constraining methods um So guys, this is it. I just want to briefly mention that At this moment, we've only considered, you know, the interpretation that comes the information that comes from the physical background alone But you can get game even more inside if you cross correlate your diffuse gamma background with others like for example, catalogs of of resolved Galaxies Or even the information related to the large-scale structure that you can extract from Irritational lens either from galaxy galaxy lensing or cnb lens And we have measurements in some case even detection for these cross correlation terms and they are complimented to what I told you What I told you about um in this presentation So, yeah, I think that's only only my conclusions lights I just want to say again that we have a new measurement of the envelope aspect in romanosanthropy For the first time There's evidence that the signal Comes from multiple components And this with this measurement with this new data you can gain more insight on our physical surface and you can put constraints on other contribution like for example Dark matter Induced gamma rays and the exclusion limit that you get are competitive With what you get from simply considering the overall intensity Of the diffuse gamma back when I forget about any So, yeah, that's it. Thank you, matthia very much Hello, yeah So it's thank you matthia for this super interesting seminar Now we should pass to the round of of questions So, please remember that you can ask questions to matthia via the google qna And the webinar webpage and via twitter as as well with the hashtag So So first I have a question here from the audience Robert I think that you have some one Yeah, I have Please go ahead. Nice. Okay. I have a question for matthia. First of all, the the webinar Thanks, I want to ask you because you presented all this constraint in terms of the annihilation or the gain is Spectrum into Be bb bar How it changed how is the dependence with respect with other channels if it is less I mean, do you require to have a sub spectra in order to make this cross correlation between different beings in energy or Or a sharp one also would make the the job so we we we checked and We considered three different possible channels bb bars 1000 years and in none of those cases we Identify features in the data that suggests So all of those three cases You still only can put constraints And the actual the actual constraints both for annihilation and decay you can find them in the paper. There are the the relative plus Okay, thank you. Uh, so in that sense is the bb bar is the most Standard, I mean the most nicer I guess because I I can check the I'm going to check the paper of course, but Uh, what do you mean by nice? Do you mean more constrained? Yeah, I think so. I think so, yeah Okay, I have one question, but maybe I can other people can ask So I also have a question matthia Nice So I'm just wondering why do you use did you use the two masks? So the the three catalog the three fgl and the two Right, right. So is that you are not trusting the three Fgl catalog? You really should consider them as two independent measurements and and indeed Since you know, which are the sources that you are Masking in the new sources that you are masking the three fgl sources different masks help a lot in Seeing the effect of those sources in the way the cp depends on the energy So you can you can relate how the two different datasets change To the effect of the search that you have masked For example, if you if you Imagine that you you're going from two fgl to three fgl your only masking sources that contribute I don't know below two gb Then you should clearly see that effect In the way the cp depends on the energy. You should only see differences below two gb Including different different cuts in sensitivity in flux Help you understand what's the effect of the sources you're masking in the signal I see thanks I I have a question Yes, um When you show the cl the measurement I think it was on page nine. So and you show the values of multiple In which you fit your Your data so from 40 to a country member for 700. Yeah, so how do you choose them? Yeah, so At low energies We introduced the sorry a lot multiples We introduced the cat to get rid of the possible contamination for the galactic foreground Can I go back to the slides? Let me share again the the screen If I manage You see it Yes So yes Look at this map over here We're masking the diffuse gamma the diffuse foreground But even outside the mask there's still some possible contamination due to the galactic foreground and however, these are large Scale feature that we only contribute at low l. So the cat at low l is introduced To get rid of these features and we check different cuts in l and and determine that The value of 49 is the lowest we can get to be sure that we are indeed cutting everything from the galactic For the galactic foreground at higher ls So here about 708 that's where the arrow bar starts to blow up because of the corrections due to the psf And and that's that's a value that I mean this doesn't matter much because in any case producing a fit To these points with a huge error bars wouldn't change much your your results of the fit by the way Since as I showed here the effect of the psf the window beam function is important specifically in particular low energies This effect the blowing up of the error bars shows up particularly at low energy So in this panel and not in this one. We could have increased the multiple range At higher energies to higher multiples. So go up going up to I don't know Couple of thousand in this case only the higher energies But we decided to work with a constant multiple range. So we decided to go for you know Yes, okay And maybe I can ask a second thing When you have there is final results, uh, the c The cl is a function of the energy For the free fgl case You have a deep Around 10 gb, right? Yeah. So you have an understanding of that or it's just you think it's just statistical effect fluctuations We think it's just a physical fluctuation And indeed so you mean you mean yeah, yeah, exactly. Yeah, the blue ones are actually There but the red ones go down There yeah, there are the ones that are effect, you know, they have a large arrow bars And we do think this is just a statistical fluctuation indeed if we change Some of the details of the analysis like for example The fact that these data points are obtained only considering uh Events that convert in the front end of the telescope But if we also include the one in the back Then these two data points go a little bit higher. So this feature this deep Is not stable by considering different Selection section in the data processing. So we do think it's just a statistical fluctuation Okay, we've also we've also played a lot played a bit with uh, passe data So that would be the even more recent reanalysis of all the the events that it by Fermi and even in that case Those two points disappear Well, spear they go higher Sorry Yes, yeah, okay Um, okay, can see a question from aron vincent So he's asking whether there's here any insight with the respect to the lethal paper about unresolved point sources Um, you mean the one on the galactic center That would be a different target, right? I don't know So if I if I if I have the correct paper in mind, they they use a different method related to the photon One point for the count distribution to analyze the regional galactic center while we are focusing on the, you know, overall distribution of the digital camera backwards. It would be interesting to do the same analysis that they do Also on larger scale of digital camera backwards Well, if I can add can I add something please? Yeah, this has been done Okay, high latitude. This is the second attempt paper actually have two papers Well, this the same kind of analysis is done not at the galactic center, but for High latitudes, so it's basically Constructing the log n log s distribution For the unresolved part of the sky, which is at high latitudes So Okay, thanks Are there more questions? I don't see I see a question from Sergio Palomares So he's asking Will you expect a shoulder from cascade of high energy photons even with a single power law or in different worlds? Is the cascading process taking into account in your results? No, it's not taking into account because our our interpretation is really just a phenomenological naive level zero first thing that you try But cascades should be included absolutely and in principle they can produce feature in your in your energy spectrum especially at lower energies since what is driving our feet are The very very one or two points below one gv Then then of course the low energy part of the spectrum should be correctly modelled. So that's a very good point We're not again Thanks So there are more questions from the audience maybe the last one. Yeah. Yeah, I have a couple more The idea in your respect number three there is this the star forming galaxy express to expect Spectrum why it is top at 10 to the 5 m e b? There is a Let me go back. Sorry This way I mean Yeah, yeah, so um, I must I must confess that there's no reason we we just Since this comes from taking the prediction from a paper The prediction from that paper was just stopping here So we we couldn't plot anything above that Without having to reproduce their results. So there's no physical reason should go down Following the same behavior. Okay. It's like a specter like a a dumping I mean, it's a it's a power low and then at some point there's the dumping due to the ebl attenuation like with Yeah Okay, so yeah a little bit with this in the same idea I mean when when you make the analysis with two broken power low, it's not could be equivalent to a power low with a with a exponential cutoff like is the phenomenological Assumption for some type of dark matter spectrum um So, you mean if the the the break that we see high energy can be due to the ebl absorption Yeah, oh, yeah any any any process that produce a exponential cutoff like you're like a dumping Yeah broken like a sharp edge, but more soft Absolutely. Absolutely. Absolutely. And then I mean the At the end, I mean these points at higher energies are the one with the largest errors. So it would be easier to fit Something that goes down in any way, right? Oh, yeah, yeah, that's true and and the other because Since all the questions are more or less correlated the The point that then when when you compare the for instance the constraints from the dwarf galaxies and this method, what is the the the The point that is restricted more the I mean Why you cannot go with stronger constraints. It's just a lack of statistics or is the the apparatus itself the The ABS of the Offering okay doesn't allow you to to go to do better than the dwarf, for instance. Yeah. Yeah um I think what is what is dominating these results is the measurement that you have you you would like to have more Smaller error bars on the on the on the cp on the tactic I don't remember our perspective so that you can be sensitive to smaller deviation with compared to the you know overall prosaion description if you can reduce Where is it? If you can reduce these error bars, then you will be able to put more string just constraint and that comes at the end of the day is relating to um Yeah, statistics statistics and possibility of maybe Extending the multiple range in order to have a longer lever arm when you a lever arm when you do the fifth um Yep, something like that. There's already been a huge improvement a huge step forward with respect to the original measurement of 2012 In that case, we had detection of power of anglo-cars spectrum Of the under of seven sigma seven sigma. Yeah, sorry five sigma only around one g And now we extended it to lower energies and the significance increased Thanks to the improvement since 2012 going to pass eight Is expecting to again be another another uh important step forward. So we should read you say Yeah, and more or less there is a time window for For that maybe next year two years two more years to I know I know that there's uh work But I I'm now involved in that so I can give you a time scale Okay, thanks. I don't know if the others has questions. So I don't see more questions So, yeah, so let's thank matthew again and all our viewers Yes, thanks a lot And we'll meet in two weeks for the next latin american webinar on physics So thanks a lot, man All right